CLEAR: Calibrated Learning for Epistemic and Aleatoric Risk
Ilia Azizi Juraj Bodik Jakob Heiss Bin Yu · 2025
Read CLEAR: Calibrated Learning for Epistemic and Aleatoric Risk by Ilia Azizi Juraj Bodik Jakob Heiss Bin Yu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-08150.txt.
Description
Ilia Azizi, Juraj Bodik, Jakob Heiss, and Bin Yu merge two kinds of doubt into honest forecasts in a 2025 paper. Predictions carry aleatoric doubt, from noisy measurement, and epistemic doubt, from scarce data. Most methods tune one and slight the other.
Their method, CLEAR, blends both with two dials tuned for true coverage. It accepts any pair of estimators for the two doubts. Tests paired quantile regression with stability-based ensembles, then deep ensembles with a joint quantile method. Across seventeen real data sets, intervals shrank by about a fifth to a quarter against single-doubt baselines while keeping nominal coverage. Gains grew largest where one doubt type dominated.
Catalog record
| CREATOR | Ilia Azizi Juraj Bodik Jakob Heiss Bin Yu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 135 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-08150